| Image information plays an important role in the development of human society. Image is the human perception and understanding of nature’s most basic media.It plays an extremely important role in people’s perception of the world correct stability. In the image acquisition process due to various uncertain factors makes the image information can not fully reproduce the real scene, including the acquisition equipment limitations, environmental impact, encoding and decoding of the quantized error factors. We must take some corresponding methods to improve the image quality for these cases. Methods for Restoring the original and improving image quality is image enhancement processing. It makes the image definition enhance also makes image colorful.As the most basic unit of a video,the basic idea of image enhancement can commence from the space domain and frequency domain. The spatial domain enhancement is based on image pixel information. The processing object is in the image pixel, pixels, and the neighborhood; The frequency domain enhancement base on Fourier transform of image. The processing object is the frequency components of the image information. To get the final image enhanced by Fourier inverse transformation. This paper mainly studies the enhancement algorithm for video products. The algorithm used in video products is greatly affected by the operational performance. So this paper first to the traditional, mature methods are introduced and summarized, with the brightness contrast and color saturation enhancement as the object of study. Based on Light and human visual characteristics, brightness information in the original image are separated. It can be divided into reflection component of light and the background component. To global contrast enhancement in the background component of smooth area, and to local contrast enhancement in the reflection component of detail texture region. To saturation global enhancement brightness related in the chrominance components.Traditional background separation is the original image with Gauss smoothing filter convolution to obtain the background component. This paper uses bilateral filtering method of multi-scale filtering image noise information. and the image after filtering is convolved with the bilateral filter. On one hand, to maintain the loss of edge information well. On the other hand, noise information is greatly reduced. in the reflection component separated. In the enhancement of background component, enhancement using histogram method of global statistical information in the traditional methods. Because of the distribution of pixel information uncertainty, it make the image enhanced the useful information reducing and peak area enhancement greatly. And the method used in this paper can keep the detail information of the image is not compressed, limit the peak area enhancement too greatly, and to correct enhance curve refer to image content changes before and after the frame. It makes the video image after processing clear and natural transition.In the reflection component processing, considering the characteristics of human visual to high light regions insensitive and dark area sensitive, to gain weight more for the details of the image in the middle of the distribution of gray scale value. To control on both ends of the weighting coefficients in the range of relatively small. At the same time to consider the correlation of a pixel to be processed and the surrounding area pixels and the correlation of the frame before and after to determine the pixel final gain.Background and reflection components processed are fused to get the final brightness enhancement of the output results. To control the gain of color saturation according to the increase in brightness, thus we get the output results. Through the comparison of simulation experiment and enhanced effect in the traditional way, enhancement effect is more obvious, natural, no color deviation, and computation accords with current hardware requirements. |
PDF Full Text Request